Search emails in Hermes's Gmail using Gmail search syntax.
AI agents call mail_search to retrieve information from Hermes Google without modifying anything — typically the context-gathering step in research, monitoring, and reporting workflows, before the agent takes action elsewhere.
This tool retrieves email data from Gmail but does not create, modify, delete, or execute actions. The search capability is inherently read-only. Severity is medium rather than low because email often contains sensitive information (personal communications, financial records, verification codes, etc.), and an AI agent could misuse search to extract private data, though the direct impact is limited to information…
From the tool's definition Tool name is 'mail_search' and description states it 'Search emails in Hermes's Gmail using Gmail search syntax.' The verb 'search' indicates data retrieval with no modification.
Attacks that exploit this kind of access
Search emails in Hermes's Gmail using Gmail search syntax. It is categorised as a Read tool in the Hermes Google MCP Server, which means it retrieves data without modifying state.
Register the Hermes Google MCP server in PolicyLayer and add a rule for mail_search: allow, deny, rate-limit, or require approval. Point your MCP client at the PolicyLayer proxy URL and the rule is enforced on every call, before it reaches Hermes Google. Nothing to install.
mail_search is a Read tool with low risk. Read-only tools are generally safe to allow by default.
Yes. Add a rate_limit block to the mail_search rule in your PolicyLayer policy. For example, setting max: 10 and window: 60 limits the tool to 10 calls per minute. Rate limits are tracked per agent session and reset automatically.
Set action: deny in the PolicyLayer policy for mail_search. The AI agent will receive a policy violation error and cannot call the tool. You can also include a reason field to explain why the tool is blocked.
mail_search is provided by the Hermes Google MCP server (jimmy-larsson/hermes-google). PolicyLayer sits as a proxy in front of this server to enforce policies before tool calls reach the server.
Every MCP server has a record like this.
Type a name, get the same breakdown: verified identity, auth posture, risk grade, capabilities, recommended policy.
Teams ship this data inside their own products. See what a licence covers →